From: Reid Wahl <nwahl@redhat.com>
To: gcc-help@gcc.gnu.org
Subject: Re: -fno-builtin not preventing __builtin___snprintf_chk in gcc 11.2.0-19ubuntu1
Date: Sun, 21 Aug 2022 19:39:40 -0700 [thread overview]
Message-ID: <CAPiuu9_=ZtFZUD+3+vzKUuvShxWezf1ngjnm0=dUmDNBQwiShg@mail.gmail.com> (raw)
In-Reply-To: <CAPiuu99E5Hodh2ES6CZj08b21k-BsLviJbPyiPJ2p5QD5bQ0Zg@mail.gmail.com>
On Sun, Aug 21, 2022 at 5:53 PM Reid Wahl <nwahl@redhat.com> wrote:
>
> Hello,
>
> I'm hitting a strange issue with GCC while trying to wrap and mock
> snprintf() for unit testing purposes.
>
> I'm setting -fno-builtin and -fno-inline, but
> __builtin___snprintf_chk() is getting called instead of snprintf().
> The net effect is that my __wrap__snprintf() never gets called. Do you
> have any advice on how to prevent object size checking builtins from
> being used, without setting a lower optimization level? I expected
> -fno-builtin to take care of it.
>
> Note that out of my whole multi-distro test bed, I've only observed
> this issue on Ubuntu 22.04, Ubuntu 20.04, and Fedora 36 Power 9. I
> don't have direct access to the test machines, but I spun up an Ubuntu
> 22.04 reproducer. It uses gcc version 11.2.0-19ubuntu1.
>
> A known working Fedora 36 x86_64 system uses gcc version 12.1.1
> 20220507. On that system, snprintf() (or more accurately
> __wrap_snprintf()) gets called as expected, instead of the builtin.
> There are also RHEL 7 machines in the testbed, so it works fine on
> older versions as well.
>
> Minimal reproducer:
> ```
> # cat test.c
> #include <stdio.h>
> void func(const char *s)
> {
> char buf[16];
> snprintf(buf, 16, "%s", s);
> printf("%s\n", buf);
> }
> int main(void)
> {
> func("hello world");
> }
>
> # gcc -g -O2 -fno-builtin -fno-inline test.c -o test
> # gdb -q ./test
> Reading symbols from ./test...
> (gdb) b 5
> Breakpoint 1 at 0x10a0: file test.c, line 9.
> (gdb) r
> Starting program: /root/git/pacemaker/test
> [Thread debugging using libthread_db enabled]
> Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
>
> Breakpoint 1, main () at test.c:9
> 9 {
> (gdb) s
> 10 func("hello world");
> (gdb)
> func (s=s@entry=0x55555555600b "hello world") at test.c:3
> 3 {
> (gdb)
> 5 snprintf(buf, 16, "%s", s);
> (gdb)
> 0x00005555555551b9 in snprintf (__fmt=<optimized out>, __n=<optimized
> out>, __s=<optimized out>) at
> /usr/include/x86_64-linux-gnu/bits/stdio2.h:71
> 71 return __builtin___snprintf_chk (__s, __n, __USE_FORTIFY_LEVEL - 1,
> ```
>
> Thank you.
>
> --
> Regards,
>
> Reid Wahl (He/Him)
> Senior Software Engineer, Red Hat
> RHEL High Availability - Pacemaker
I figured out the Ubuntu issue. `man gcc` under `-O2`:
NOTE: In Ubuntu 8.10 and later versions,
-D_FORTIFY_SOURCE=2 is set by default, and is activated when -O is set
to 2 or higher. This enables additional compile-time and run-time
checks for
several libc functions. To disable, specify either
-U_FORTIFY_SOURCE or -D_FORTIFY_SOURCE=0.
I have no idea what's causing the unit test to fail on Fedora 36 Power
9 LE (the only system in the test bed that's still failing), but I
don't have a reproducer or enough information to ask a question. I'll
try to get one set up this week and go from there.
--
Regards,
Reid Wahl (He/Him)
Senior Software Engineer, Red Hat
RHEL High Availability - Pacemaker
next prev parent reply other threads:[~2022-08-22 2:39 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-08-22 0:53 Reid Wahl
2022-08-22 2:39 ` Reid Wahl [this message]
2022-08-22 3:03 ` Xi Ruoyao
2022-08-22 4:07 ` Reid Wahl
2022-08-22 8:41 ` Florian Weimer
2022-08-22 8:54 ` Reid Wahl
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CAPiuu9_=ZtFZUD+3+vzKUuvShxWezf1ngjnm0=dUmDNBQwiShg@mail.gmail.com' \
--to=nwahl@redhat.com \
--cc=gcc-help@gcc.gnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).